The nature of neuronal words and language

doi:10.4236/ns.2010.23032

Paper Menu >>

Journal Menu >>

Vol.2, No.3, 205-211 (2010) Natural Science

http://dx.doi.org/10.4236/ns.2010.23032

The nature of neuronal words and language

Morris Henry Baslow

Nathan S. Kline Institute for Psychiatric Research, Center for Neurochemistry, New York, USA; [email protected]

Received 5 December 2009; revised 13 January 2010; accepted 30 January 2010.

ABSTRACT

Individual neurons in the brain possess natural

language in the form of energy-dependent ac-

tion potentials or spikes (S) operating on a mil-

lisecond timescale that, along with pauses (P)

between spikes, constitute a two letter (S, P)

“alphabet” that is used to generate meaningful

frequency-encoded neuronal “words”. These

words are then used to transmit information to

other neurons in the form of phrases consisting

of two or more words that are contained within

longer pause-delineated structured declarative

sentences. In this article, the nature of neuronal

words and language are described, and exam-

ples provided that illustrate the way in which

neuronal language is used by the brain to in-

teract with and interpret both its internal and

external environments. It is hoped that a fuller

understanding of the language used by neurons

to communicate may lead to development of

novel treatments for a number of human neu-

ropathies.

Keywords: Biosemiotics; Brain; Cognition;

Language; Learning; Memory; Neurons

1. INTRODUCTION

Semiotics is broadly defined as a study of signs, mean-

ings and their significance in communication, and bio-

semiotics as a study that is based on the general assertion

that all living organisms including cells are semiotic

systems. While information processing in the brain is

highly complex, each neuron uses a simple code mecha-

nism for transmitting information. This is in the form of

electrophysiological action potentials or spikes (S) of

about a 1 millisecond (ms) duration that, along with

pauses (P) between spikes constitute a two letter “al-

phabet” that generates meaningful frequency-encoded

signals or neuronal S/P “words” [1]. The term “word”

for neuronal activity in spikes/s has previously been used

to describe the electrophysiological activity of the visual

interneuron of the blowfly where each action potential

(AP) was considered a “short” word and trains of AP’s

considered “longer” words [2]. In this article a neuronal

word is defined as a single AP together with the total

pause before the next AP; a phrase is defined as a group

of two or more words, and a sentence as a temporal

grouping of such words containing a subject, and a

predicate that expresses what is stated about the subject

[1]. All neuron codes, whether continuous trains of

spikes that are characteristic of information transmitted

at low frequencies, or bursts of spikes that are character-

istic of information transmitted at high frequencies, are

made up of interactions between spikes and pauses.

There are two theories regarding how these spikes carry

encoded information [3]. One is the “spike rate code”

which suggests that information is carried in the average

rate at which the neuron fires, and that the timing of each

spike is random. The other is the “spike timing code”

that suggests that specific information is carried not only

in the average spike rate at which a neuron fires, but also

in the precise timings between each spike. In this article,

it is proposed that neurons transmit meaningful informa-

tion not only in their spike rate and spike pause timing,

but also in the spike duration or length (SL) and its rela-

tionship to the pause length (PL) between spikes. It is

this relationship that constitutes a neuronal “S/P word”,

and adds an additional dimension to the potential rich-

ness of communication between neurons.

2. DISCUSSIONS

2.1 Anatomy of a Neuronal Word

The communication method used by the brain is gener-

ated at the cellular level where individual neurons trans-

mit information to one another by generation a series of

wave-like depolarizations. This depolarization’s start at

the plasma membrane of their cell bodies (soma) and

pass along the plasma membranes of axon hillocks, ax-

ons and pre-synaptic dendrites. Each wave-like spike

consists of two components, first, a rapid depolarization,

involving the efflux of K+ and influx of Na+ at the

plasma membrane down their respective intracellular/

extra cellular gradients, a process that requires little or

M. H. Baslow / Natural Science 2 (2010) 205-211

206

no additional energy. This is followed by a re-polariza-

tion process that occurs within an absolute refractory

period during which time the intracellular negative

membrane K+/Na+ potential is restored using energy

derived from intracellular stores of adenosine triphos-

phate (ATP) via a plasma membrane surface enzyme,

Na+/K+ ATPase, in order to transport these ions against

their respective gradients [4]. In single afferents of hu-

man mechanoreceptive sensory nerve fibers of the skin

of the hand, spike times-to-peak were 0.12-0.45 ms with

a mean of 0.21 ms [5]. While the depolarization process

is pre-energized and trigger-ready, re-polarization is a

slower enzyme-dependent process that is subject to rate

regulation by all factors that can affect enzyme activity.

These include enzyme, substrate, and end-product con-

centrations, focal ionic makeup, pH, temperature as well

as the presence of competitive and non-competitive in-

hibitors. All of these factors can affect the rate of

re-polarization and the length of the absolute refractory

period which in this study varied from 0.31-0.75 ms with

a mean of 0.50 ms indicating that the energy consuming

re-polarization period was 2.38 times longer than the

depolarization period such that the length of the

re-polarization period dominated the total spike period.

In this study, the average length of a spike including the

depolarization and longer re-polarization periods in the

sensory nerve fibers was 0.71 ms with a range from

0.43-1.20 ms. The relationship between the depolariza-

tion and re-polarization periods of a single spike is illus-

trated in Figure 1. Since the depolarization is pre-primed

and trigger ready, and the re-polarization period requires

energy to reestablish ionic gradients, re-polarization is

therefore subject to cellular control. Thus, the SL can be

varied to incorporate information that is transmitted in

the form of a SL/PL ratio over and above information that

transmitted by Hz and/or specific peak-to-peak in-

ter-spike intervals. Data from [5].

During the spiking process which includes the absolute

refractory period, and until the ionic gradients are rees-

tablished, a second spike cannot be generated. In addition,

following the absolute refractory period there is a relative

refractory period where only a strong stimulus can gener-

ate a second AP, and following this period there is a pe-

riod of longer pause with no electrophysiological activity.

A spike including its absolute refractory period may vary,

but an average spike-absolute refractive period length is

about 1 ms, and is followed by a relative refractory pe-

riod of about 2 ms in duration [6]. This in turn is fol-

lowed by a pause of varying lengths before the next spike

is generated. Thus, there are two phases of signal genera-

tion, each consisting of two subcomponents. The total

spike length (SL) is made up of the sum of the depolariza-

tion and absolute refractory (re-polarization) periods, and

the total pause length (PL) is made up of the relative re-

fractory period and the pause until the next spike is initi-

ated. The total pause length can vary from 0 to approxi-

mately 10,000 ms under normal physiological conditions.

Based on these neuronal spike and pause components, a

single neuronal S/P word has been defined as the sum of

its total spike length and its total pause length in ms, and

different words are characterized by differences in the

ratio of its SL to its PL [1].

The spike-generation process that results in S/P words

is also is metabolically costly, calculated at 2.2 × 109

ATP molecules per spike [7], which requires that ATP

supplies be constantly replenished or the timing of the

spike-refractory periods will be altered, and meaningful

frequency-encoded information lost [1]. An illustration

of the components of a neuronal word as described is

graphically presented in Figure 2.

Figure 1. Cartoon showing the relative depolarization and repolarization

times of a neuronal spike.

M. H. Baslow / Natural Science 2 (2010) 205-211

207

Figure 2. Cartoon showing the components of the neuronal word, S1P7。

In this illustration, each neuronal word (S1P7 ) consists

of a depolarizing spike (0.5 ms), an absolute refractory

period when the plasma membrane is being re-polarized

using ATP (0.5 ms) for a total spike length (SL) of 1.0 ms;

a relative refractory period where the neuron can be ac-

tivated but only by a strong stimulus (2.0 ms); and a

period of inactivity (5.0 ms). The spike frequency is

(1000 ms/8 ms) or 125 Hertz (Hz). Unless the pause is

terminated by a second spike within a physiologically

relevant period, the word is incomplete and without any

quantitative meaning. Thus, a meaningful phrase must

consist of at least two words, and a sentence of at least

one phrase. The word S1P7 presented in this cartoon

reflects the S/P message generated upon exposure of the

isolated salt receptor of the blowfly to a 3 molar NaCl

solution [8].

2.2. Conversion of Frequency in Hz to S/P

Words

As shown in Figure 2, information contained in Hz for-

mat differs from that in S/P word format in that in Hz,

the signal is a measure of the time between two spikes

that is initiated at the peak of the first spike and is con-

stituted by the re-polarization period of that spike, its

relative refractory period, the pause until initiation of a

second spike and the depolarization period of a second

spike. In essence, the signal recorded in Hz splits a spike

into its two components such that a spike in Hz is a

composite formed from elements of two AP’s. On the

other hand, an S/P word as defined includes the depo-

larization and re-polarization components of a single

spike, along with the relative refractory period and the

pause until the next spike is initiated. However, the two

formats are clearly related, and while Hz is a commonly

used method of recording and describing neuronal activ-

ity, a reasonable estimation of the length of a neuronal

word can also be derived from Hz. Thus, from Hz, the

total time in ms of an average S+P word can be deter-

mined: Total time per word: (SL + PL)ms = 1000/Hz.

At a frequency of 100 Hz, the mean S+P word length

would be 10 ms. However, the total value in ms for an S

+ P word does not provide sufficient information to con-

struct a specific neuronal S/P word. For this, one re-

quires in addition either the total spike length (SL) or

total pause length (PL). Thus, for a SL of 0.5 at 100 Hz,

the S/P word would be S0.5/P9.5, for an SL of 1.0, the S/P

word would be S1.0/P9.0, and for a SL of 1.5, the S/P word

would be S1.5/P8.5, representing three different S/P words

out of a great number of possible S/P words that are dif-

ferentiated by the relative lengths of their spikes and

pauses. It is important to note that all of these S/P words

are different and may transmit different meanings, but

that all are being communicated at the same frequency

of 100 Hz. Clearly, a measure of frequency alone for

neuronal communications cannot be physiologically

accurate. When afferent neuronal messages are con-

verted from electrophysiological to chemical and back

into electrophysiological messages at synaptic interfaces

for transmission to efferent neurons, it is highly likely

that differences in the SL/PL ratios are important ele-

ments in modulation of the messages at these interfaces.

Since much of the published neuronal response data is

presented in the form of Hz, and while less accurate than

actual S/P words, in the absence of a measured SL, and

by using an average SL of 1.0, one can still derive useful

information based on Hz about how neurons communi-

cate with one another concerning the nature of their en-

vironment.

2.3. Differences in Spike Rate, Spike Timing

and Neuronal Word Codes in

Information Transfer

The spike rate code hypothesis proposes that information

is processed in the form of the average time between

M. H. Baslow / Natural Science 2 (2010) 205-211

208

spikes, and the spike timing code hypothesis proposes

that information can also be contained in the differences

in pauses between individual spikes, an analysis that can

be used to identify harmonics and periodic oscillations in

spike trains. Both of these hypotheses treat a spike as a

point source and both are valid interpretations of re-

corded data. However, these hypotheses are mirror im-

ages of one another in that when viewed at the level of

only two spikes, as presented in Figure 1, it is clear that

the time between individual spikes treated as point

sources, and the pause time between spikes will always

be the same. On the other hand, the neuronal word code

hypothesis as detailed above takes into consideration

that individual spikes have a significant time dimension

and shows that even at the level of only two spikes, there

can be an enormous number of specific S/P words that

can be generated and used to transmit different kinds of

information. This is because each neuronal word is

formed by individual components of the AP process as-

sociated with spike generation, and differences in each

component can alter the ratio between the total spike

length and the total pause length. For the three examples

provided above, S0.5/P 9.5, S1.0/P9.0 and S1.5/P8.5, their S/P

ratios are 0.052, 0.111 and 0.176 respectively, and all are

capable of being transmitted at the same frequency (Hz)

where both the inter-spike time and pause time are the

same. While all three hypotheses reflect frequency-based

neuronal codes, the neuronal word code with its spike

time dimension is the most physiologically realistic

since it is the only one that identifies a mechanism

whereby the AP can interact with the rate of depolariza-

tion-induced release of neurotransmitters from pre-

synaptic vesicles. Thus, only the neuronal word code

appears to have the potential for an enhanced richness in

information transfer, especially at the low frequencies

and short periods dictated by ongoing bioenergetic and

temporal physiological constraints. Moreover, measure-

ment of the ratios that comprise S/P words may provide

a new investigational tool, not available using the other

“codes”, with which to assess differences in neuronal

function between normal and pathological conditions.

2.4. Nature of Neuronal Language

There are three components to neuronal language. First,

neurons are “wired” in the sense that each neuron in the

corpus occupies a specific place on a brain “map” of a

given organism. Second, by virtue of its phenotype, each

neuron sends specific kinds of messages for interpreta-

tion within the CNS network. For environmental sensing

neurons, these may identify the qualitative nature of the

messages being communicated such as where in the

brain or corpus it originates, and whether the message is

a general call for stored information, or whether it is

from a specific region associated with some specific

factor such as sound, light or availability of “glucose”

(Glc). Lastly, the messages sent by neurons must also

concern the quantitative nature of sensed information

such as the importance of message itself, or the specific

levels of light, sound, pain, pressure, stretch, or sub-

stance concentrations sensed. It is this last component,

the quantitative nature of information within the brain,

where specific frequency-encoded neuronal words ap-

pear to play an important role.

As an example, based on a recorded neuronal fre-

quency of 100 Hz, a neuronal spike with a total spike

length of 1 ms and a total pause length of 9 ms would

constitute the neuronal S/P word ( S1P9) of 10 ms dura-

tion. When this word is repeated, it becomes a phrase. In

the blowfly, this phrase corresponds to immersion of its

isolated salt receptor in 2.0 M NaCl [8]. Based on using

a value for total SL of 1 ms, the specific words in “fly-

speak” for immersion in additional NaCl concentrations

of 0.5 and 1.0 M are S1P18 (52 Hz) and S1P12 (80 Hz)

respectively. In addition, if the SL is known along with

the total S + P time in ms derived from Hz, the total

“pause” time in ms can be derived using the following

general relationship: P = [1000 ms – (Hz × SL)]/Hz.

It has also been proposed, that the language of neu-

rons can be written in any of a number of fre-

quency-encoded formats including Hz, S/P and musical

notations and can be translated into any of the oral, writ-

ten or symbolic human languages as well [1]. Impor-

tantly, using artificial electronically generated frequen-

cies that mimic the frequencies of recorded natural neu-

ronal words, it has been demonstrated that it is possible

not only to record these neuronal words, but also to

communicate directly with the central nervous system

(CNS) of rats. Where this has been done, the manufac-

tured words have been shown to elicit the same behav-

ioral responses in rats that would have been generated by

exposure to a specific environmental stimulus [9].

2.5. The Dictionary of Neuronal Words and

their Representational Nature

The dictionary of neuronal S/P words in an ms time-

frame is relatively short, being limited at its upper level

by the absolute and relative neuronal refractory periods

and at its lower level by physiological requirements such

as maintenance of the human heart rate at 68 beats/min

(1.13 Hz). With the average spike including an absolute

refractory period of 1 ms duration followed by a relative

refractory period of about 2 ms, the highest possible

frequency would be S1P2 or 333 Hz. If the system were

driven by a strong enough input eliminating the relative

refractory period altogether, the highest possible fre-

quency might be S1P0 or 1,000 Hz.

Since the brain’s primary form of communication,

both internally and with the outside world, is in the form

of these electrophysiological messages, it follows that

neuronal words must also be highly representational in

nature. For example, audible sound in many animals

may range up to 20,000 Hz and visible light is in the

M. H. Baslow / Natural Science 2 (2010) 205-211

209

Figure 3. A “dark room” paradigm of brain function.

Table 1. List of several neuronal S/P words and their meanings.

Organism Sensor Word Meaning Reference

Blowfly

sensillum salt S

1P499 NaCl 0.10 M [8]

1P45 0.20

1P27 0.25

1P18 0.50

1P12 1.00

1P9 2.00

1P7 3.00

Ground beetle

Sensillum salt S

1P1000 NaCl 0.001 M [11]

1P71 0.01

1P29 0.10

1P23 1.00

Hamster

taste cells acid S

1P286 pH 2.5 [12]

1P312 3.0

1P1999 4.3

1P4000 5.0

Cat brain

hypothalamus glucose

glucosensing steady-state S

1P10000 normal Glc level [13]

Glc i.v. 30 min S

1P2000 elevated Glc level

Rat

taste cells quinine S

1P21 quinine 0.01 mM [9]

Rat brain

hippocampal sound S1P499 frequency 2 kHz [10]

neuron S

1P142 10 kHz

Guinea pig brain

GABAergic H

2O2 S

1P53 H

2O2 0.0 mM [14]

neurons S

1P40 1.5

range of 1014 Hz. For the brain to comprehend its exter-

nal environment, these frequencies must first be con-

verted into representations of those wavelengths at less

than 1000 Hz, with the same meanings as the original

signals, which can then be transmitted by neurons within

the CNS network. These representations can also be

“memorized” in the form of structural and biochemical

changes in the dendrite-synapse-dendrite (DSD) fields

M. H. Baslow / Natural Science 2 (2010) 205-211

210

between neurons, and later recalled to “imagine” the

nature of the original wavelengths even without external

input. For example, one can imagine “in the minds eye”

the color “red” with the eyes closed, and with eyes open

can also compare a newly observed representation of red

with that which was previously seen and stored in mem-

ory. This phenomenon is illustrated in Figure 3 where a

specific sound tone of 2 kHz impinging on the ear is

converted into a representational neuronal word at 2 Hz

for CNS network processing.

In this illustration, the translation by a sense organ of

the specific frequency-encoded environmental informa-

tion for the approximate musical tone B6 (1975.53 Hz) is

illustrated. The sound sense organ first converts the tone

into a brain frequency-encoded S/P word representation

for B6 at 2 Hz (S1P499) that is then compared within the

CNS network to auditory area DSD memory constructs.

During this comparison related historical documents in

other brain areas may also be probed for relevant infor-

mation. Upon an executive command, appropriate mus-

cles can also be activated to reproduce the 1975.53 Hz

frequency, exporting it to another organism and also

sending it back to the sense organ where it can be com-

pared to the original tone. Importantly, the S/P word for

the tone B6 or any other musical tone can also be synthe-

sized de novo by the brain from memory engrams with-

out external input and can be “heard” within the brain as

recalled sounds or melodies. All of these functions occur

in a ms/sub-ms timeframe and apply to any other sensed

parameter, whether derived internally or external to the

brain. The specific auditory data presented in Figure 3 is

adapted from rat experiments [10].

2.6. Examples of Neuronal Words and their

Meanings

In Table 1 descriptions of some S/P words and their

meanings are presented. In each of these cases, the S/P

words have been derived from frequency data by using an

average SL including an absolute refractory period of 1.0

ms, and rounded to the nearest integer. However, it is also

apparent that neurons can operate in a sub-millisecond

timeframe [10], which is in keeping with the microsecond

timeframe dictated by physiological demands and com-

munication networking requirements of the brain.

3. CONCLUSIONS

In this paper it is proposed that neurons possess natural

language in the form of specific neuronal words, phrases

and sentences, and evidence is provided that when spe-

cific neuronal words are known, it is possible to mimic

them and generate electronic words to communicate

intelligently with the CNS. While most investigative

recordings of neuronal activity are in the form of Hz, the

differences between Hz and word formats have been

described, and a case has been made that the word for-

mat is the only format that can be physiologically asso-

ciated with the release of specific quanta of neurotrans-

mitters at synapses and is therefore more representative

of the potential richness of information that can be

transmitted by neurons. In addition, based on physio-

logical limitations associated with the bioenergetic na-

ture of neuronal AP’s, it has been shown that much of

the information about the external environment that is

transmitted to the brain must first be reduced to a fre-

quency-encoded representation of that information that

can be communicated at < 1000 Hz. These representa-

tions can be acted upon immediately, memorized when

appropriate, and the memorized information recalled as

needed in the form of such representations within the

brain. Of note, in humans and other animals these inter-

nal representations can also be converted back into their

original electromagnetic spectrum frequencies by exter-

nally re-creating the original environmental source such

as by reproducing a particular memorized sound fre-

quency. Finally, it is hoped that this paper will encourage

the continued analysis of the nature of neuronal words

and language, and that such studies will lead to addi-

tional development of novel electronic methods for

communicating directly and intelligently with the CNS

as has already been attempted in treatment of several

human neuropathies [15]. With such specific fre-

quency-encoded words as tools, it may be possible to

alter or induce new behaviors, and thus lead to success-

ful treatment stratagems for a variety of human psycho-

logical and neurological conditions.

REFERENCES

[1] Baslow, M.H. (2009) The languages of neurons; An

analysis of coding mechanisms by which neurons com-

municate, learn and store information. Entropy, 11,

782-797.

[2] Borst, A. (2003) Noise, not stimulus entropy, determines

neuronal information rate. Journal of Computational

Neuroscence, 14, 23-31.

[3] Clifford, C.W. and Ibbotson, M.R. (2009) Response vari-

ability and information transfer in directional neurons of

the mammalian horizontal optokinetic system. Visual

Neuroscience, 17, 207-215.

[4] Baslow, M.H. and Guilfoyle, D.N. (2007) Using proton

magnetic resonance imaging and spectroscopy to under-

stand brain activation. Brain and Language, 102,

153-164.

[5] Brink, E.E. and Mackel, R.G. (1993) Time course of

action potentials recorded from single human afferents.

Brain, 116, 415-432.

[6] Berry II, M.J. and Meister M. (1998) Refractoriness and

neural precision. Journal of Neuroscience, 18, 2200-

2211.

[7] Lennie, P. (2003) The cost of cortical computation. Cur-

rent Biology, 13, 493-497.

M. H. Baslow / Natural Science 2 (2010) 205-211

211

[8] Gillary, H.L. (1966) Stimulation of the salt receptor of

the blowfly. II. Temperature. Journal of General Physi-

ology, 50 351-357.

[9] Di Lorenzo, P.M., Leshchinskiy, S., Moroney, D.N. and

Ozdoba, J.M. (2009) Making time count: Functional evi-

dence for temporal coding of taste sensation. Behavioral

Neuroscience, 123, 14-25.

[10] Takahashi, S. and Sakurai, Y. (2009) Sub-millisecond

firing synchrony of closely neighboring pyramidal neu-

rons in hippocampal CA1 of rats during delayed non-

matching to sample task. Frontiers in Neural Circuits,

3(9)

[11] Merivee, E., Renou, M., Mand, M., Luik, A., Heidemaa,

M. and Ploomi, A. (2004) Electrophysiological responses

to salts from antennal chaetoid taste sensilla of the

ground beetle Pterostichus aethiops. Journal of Insect

Physiology, 50, 1001-1013.

[12] Gilbertson, T.A., Avenet, P., Kinnamon, S.C. and Roper,

S.D. (1992) Proton currents through amiloride-sensitive

Na channels in hamster taste cells. Journal of General

Physiology, 100, 803-824.

[13] Anand, B.K., Chhina, G.S., Sharma, K.N., Dua, S. and

Singh, B. (1964) Activity of single neurons in the hypo-

thamic feeding centers: effect of glucose. American

Journal of Physiology, 207, 1146-1154.

[14] Lee, C.R. and Rice, M.E. (2008) Hydrogen peroxide

increases the excitability of substantia nigra pars reticu-

lata GABAergic neurons. Society for Neuroscience

Meeting, November 16, Program 179.2, Poster # QQ37.

[15] George, M.S. and Aston-Jones, G. Noninvasive (2010)

techniques for probing neurocircuitry and treating illness:

Vagus nerve stimulation (VNS), transcranial magnetic

stimulation (TMS) and transcranial direct current stimulation

(tDCS). Neuropsychopharmacology, 35, 301-316.